We've all been there, screaming at the TV when a receiver drops a pass on a 3rd down, in the endzone, or heaven forbid in a possible game-winning situation. It could easily be argued that some drops are more costly than others (even if it's unfair to blame any particular player based on such a small sample size) - hell, I would love to have play-by-play data for drops to be able to incorporate them into Advanced NFL Stat's Win Probability Added metric (to know which drops lowered a team's chance to win the most, and which receivers lowest WPA for their combined drops).
What I would like to argue, is that on the whole (when examining a large sample size), drop-% (defined as drops divided by receptions+drops, x100) is not correlated to a WR's overall performance. That a WR does not need a good/low drop% to be a top tier WR. I casually made this observation while going over some data from Pro Football Focus. For instance, Wes Welker constantly flirts with (and often exceeds) 10% dropped passes (which would have ranked 51st in 2012). Calvin Johnson had a drop-% higher than 10% in his record breaking 2012 season. That is not to say it's not a nice bonus, that if all things were equal it would be a good tie-breaker between players, but that if any catching-related ability is valued it would be the ability to make "athletic" catches so that your QB can put it in a spot only you can get it (as Jordy Nelson displays nicely here). I argue that a "poor" drop% does not detract from a talented WR.
Note: I do acknowledge that the game is played in small-sample sizes - that a drop or 2 can kill a game if it happens at the wrong time, but what I argue that it is more important to have the ability to consistently get open & create opportunities for your QB to get you the ball. You may have 2 drops in 1 game, but you will have zero in many others & be gaining yards & first downs for your team.
Now we've reached the statistical portion of the evening...Wait! Don't run! I'll try to keep it brief with the numbers, but I will gladly share the spreadsheet I worked on upon request. To judge WR "performance" (aka: the end result), I used PFF's passing-grades. I do not know exactly what goes into them, but I do know they do their best to take everything into context (down/distance, time left, score, etc etc). So I trust that the context of drops are taken into account as well. All of the numbers I used come from PFF, with the exception of 1st-downs (which include TDs) which I found on Pro Football Reference. All numbers I used were from the 2012 regular season. I examined 79 WR who had at least 30 receptions (and met PFF's 25% snap-count requirement for routes-ran). I pro-rated all the receiving grades to 16 games.
I plotted the drop-%, along with 6 other statistics, of each WR against his pro-rated receiving grade. I then calculated the R-squared value, which indicates the level of correlation between 2 groups of numbers. This value is always positive (which one expects when a number is squared), and the closer to 1.0, the closer the groups are correlated. (For example, a the wins & losses of teams (ignoring ties) would have a perfect 1.0 correlation (as someone with 4 wins must have 12 losses, etc)).
The other statistics I plotted against grade were as follows: Targets/game, Targ% (of routes run), Receptions/game, Rec% (of routes run), Yards-Per-Route-Run (YPRR), and % of routes run that resulted in 1st downs or TDs (1D%). From lowest to highest correlation, here are the R-squared values for each of the stats vs. receiving grade:
Drop% .19 (correlated negatively, higher drop% with lower grade)
As you can see, drop% is more more weakly correlated to WR success than any of the other metrics that related to their ability to do positive (rather than one highlighting their inability). Obviously the reception-related numbers are more highly correlated than the target ones, as not only are they presenting themselves as a target, they're also catching the ball & creating more value. Next, YPRR and 1D% are even more strongly correlated because not only are they getting open & making the catch, they are producing once they have the ball.
In addition, here are the R-squared values for Targ% & Rec% vs YPRR & 1D%
Once again, we see that WR success starts with getting open and is furthered by doing something positive once you receive the ball. One guy may rarely drop the ball, but if he doesn't get open often** he doesn't present much value to his team. On the opposite hand, another receiver may have a non-ideal drop%, but if he's talented he will offer many more chances for his QB to get him the ball.
**I will acknowledge, that if a player is obviously not a "top talent" (perhaps your 2nd or 3rd WR), you would want him to have a lower drop% so that he does not blow the few opportunities he ceates for himself.
In addition, I found that drops accounted for only 16.4% of all incomplete passes these WRs faced (and the median inc-% was only 15%). This further demonstrates that a WR need not catch 95%+ of the catchable balls thrown his way to be a quality WR, he only need get open as often as possible and hope that he has an accurate QB to take advantage of this. He may drop a pass here or there, but if he is good enough to get the ball thrown his way often - good things will happen.